As far as I understand, very young children have no trouble distinguishing many different phonemes. However, when they get older they lose the ability to hear the difference between different phonemes that their native language doesn't distinguish. As a result, German native speakers generally can't distinguish between "believe" and "belief" or "cap" and "cab".

At what age does this ability get lost?

  • Is it actually true that German native speakers cannot hear the difference between "believe" and "belief"? Could someone please supply a reference?
    – Greg Lee
    Commented Sep 27, 2017 at 1:17
  • Curiously, it seems to be yet more complicated. In my experience, G speakers tend to believe they can hear a difference b/w G final voiced & unvoiced obstruents, yet they often fail to clearly pronounce them even when speaking eg English which has the distinction. Hence the G-colored [lɛts kɔl ə kɛp] for E ...[kæb]; 'cab' ~ 'cap' get conflated. Commented Sep 27, 2017 at 13:49
  • Every person can detect/hear the difference between two distinct phones when pronounced in isolation. It is only when the sounds come up in context that the native tongue filters 'similar' sounds. German speakers can hear the difference between 'belief' and 'believe' or 'cap' and 'cab' (even '*kep', if that were a word) if pronounced in isolation. What they cannot do is believe (for want of a better word) how such a 'small' distinction can be used in a language to impart contrasts. Commented Sep 29, 2017 at 17:53

2 Answers 2


Infants can reliability perceive contrasts between sounds in various languages. However, by the age of 10-12 months, babies' ability to distinguish between contrasts important for their native language(s) continues to improve while the ability to hear non-native contrasts declines. In other words, the brain and auditory systems "tune into" aspects of the incoming input that are important for their native language and starts "ignoring" variation that might be important in other languages but not in the native language(s). See, for example, https://www.ted.com/talks/patricia_kuhl_the_linguistic_genius_of_babies#t-227277

Now, for older listeners, like adults, there is variation in how well learners are able to still "hear" these forgotten contrasts or re-learn them. Tests of language aptitude may attempt to measure this ability. For example, http://www.lingref.com/cpp/slrf/2008/paper2382.pdf.

In addition, re-learning may be easier for people who have overheard the to-be-learned language in childhood, even if they don't remember anything about it consciously. See, for example: http://citeseerx.ist.psu.edu/viewdoc/download?doi=


I don't think there is a general age. Some people retain the ability into old age; some people retain it into their 30's, maybe it declines for some around 20. Since p and b, f and v are contrastive phonemes of German, very few German speakers lose the ability to distinguish those phonemes. There might be a question about German speakers detecting it in word-final position, though.

English speakers who don't speak Logoori don't generally have the ability to distinguish i ɪ; u ʊ in that language, even though English has that contrast. The reason is that those vowels in Logoori are much closer together than they are in English.

There is a fundamental methodological problem with answering the question, that there is no good way to test if a person can "distinguish" two sounds. You would have to present listeners with two recordings and ask if they are different or the same. Assuming these are sounds of a language that they don't know, they don't have any idea about phonemic boundaries in that language, so you're effectively asking them to set aside their own language, and listen for absolute acoustic differences. You could perform the test by giving similar words, different tokens of the same word, or playing the exact same token twice – the question would be "are these the same, or different". I don't know of any good studies looking at sampled speech that way, though I've done this with pure tones with marginally different physical properties.

If someone does the experiment, they will have to include children in the subject pool, because we don't know how children (say age 8) are perform on this test. We know that ultimately they could learn any language, but this kind of experiment is not the same as learning a language.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.